Here is the poll for today. Indicate what you think, then read what I think.
Reporting the outputs of statistical analysis is tricky. It is not easy to be both correct and readable. There are so many nuances that can make a seemingly correct statement incorrect. For example when reporting the R-sq value from a regression analysis, I have seen it written that “temperature explains 63% of sales.” This is almost correct, but should read “temperature explains 63% of the variation in sales.” There is a subtle but important difference (which can be the subject of another post!)
Some reports were less than ideal
For many years I asked students to write up their results and then was frustrated when their reports were incomplete, incorrect, imprecise, incomprehensible or all of the above. It then struck me that I was asking them to do something that I hadn’t taught them to do. It was like asking them to assemble a complicated piece of furniture, giving them only the box of parts, with no instructions, then complaining that the end result was poorly constructed.
So now I give them a pattern. It comes in the form of a set of questions in the Learning Management System, where they are required to fill out the template after performing the analysis using Excel. There are several sets of data for the students to practice on. From it they gain immediate feedback.
A question looks like this:
An example of the type of question we use to teach how to write up a regression analysis.
The dropdown boxes provide alternatives, and numbers are entered in the open boxes.
An example of the alternatives provided
In the final assessment students are given data to analyse using Excel, and they write up the results without prompts.
The students often memorise the template, and use it to write a correct, clear, complete, comprehensible report. The standard of write-up has improved markedly, and I no longer get frustrated at reading poor reports in the final assessment. Some students still manage to avoid learning and write non-sensical sentences, but the proportion has reduced considerably.
So am I right? I believe I am. Even if the students don’t completely understand what they are doing, they can at least produce something correct. They are aware of statements that are incorrect, as they were penalised for them in the practice exercises. How many of us really do understand all about techniques we use, especially at the start? Templates give students a correct recipe to work from so that they build on a sound foundation. If there were enough time in a course, you could get the students to work out a template as a class, or individually, giving them feedback as they go.
Please let me know if you disagree or have concerns about the use of templates. It’s really good to have opposing views aired. And maybe you will convince me to abandon my templates!
For students of statistics reading this, you can find my recommended pattern for writing up a regression in the YouTube video on regression.